we are only just realising how much was achieved on these lines in ancient times with 

 no more rain than now falls and later destroyed. This is a matter for engineers and 

 geologists. 



V/ind Erosion Deserts. Sand Surfaces 



As one approaches the cores of the great desert regions and rainfall becomes less 

 and less frequent, a stage is reached at which erosion by wind has for ages exceeded 

 that by water. The landscape becomes lunar. Stony plateaux alternate with escarp- 

 ments, isolated hills, gentle isolated depressions and wide sand- covered plains. 

 Drainage lines fade out and disappear. Since concentration of run -off may be negli- 

 gible, variation in the available moisture from place to place now depends on local in- 

 creases in elevation which attract more frequent rain, and on the capacity of the gen- 

 eral surface soil to absorb and retain rain where it falls. 



When the soil is suitable in this respect temporary grazing springing up from dor- 

 mant seeds is able to mature and seed itself from a single rain storm after several 

 years of drought. A limit is probably set to the period by the viability of the seed and 

 its physical durability under conditions of sand blast and strong solar radiation. The 

 best soil is undoubtedly blown sand which is relatively clean of fine dust particles 

 between the grains. Water can descend very quickly through this soil since its anti- 

 wetting property is low and its permeability high. Owing to capillary tension a given 

 charge of water applied at the surface of dry sand will sink to a certain depth and no 

 more, the depth being something of the order of eight times the immediate precipitation. 

 Water which has reached a depth of 20 to 30 cm. remains as a moist unsaturated zone 

 for several years because, sand being a very poor conductor of heat, the temperature 

 is constant and 'breathing' nil. The sand both above and below is dry. A sand accu- 

 mulation is a good desert soil for two other reasons. In a wind erosion plain it is the 

 only place where deposition can balance removal; and it produces the only sloping 

 surfaces capable of appreciable local run- off concentration. Hence one finds that the 

 most favourable vegetation sites, indeed the only sites, lie along the lower gentle 

 slopes of dunes. 



I suggest that since blown sand has an almost constant composition and texture 

 it might well be used as a standard soil for the purpose of estimating from the presence 

 or absence of vegetation the mean useful rainfall of those areas for which no long- 

 period records are available. 



Variability of Rainfall ftom Year to Year. Unreliability of Records 



In temperate climates where adequate rain falls throughout the year we are accus- 

 tomed to some deficit in any year from the mean annual value, and we do not bother 

 about it. But as the mean annual value decreases towards a desert region, and be - 

 comes confined to one season only, the expected deficit does not obligingly diminish 

 in proportion. A stage is reached at which the probable deficit at any one place is of 

 the same order as the mean annual rainfall. Beyond this stage the probable rainless 

 period exceeds one year. Instead of asking 'have the rains been good this year?' we 



